8 results about "Defragmentation" patented technology

Methods, systems, and devices for increased file loads using file based optimization (FBO) and prefetch within a memory system are described. A system may support using a prefetch command to prioritize prefetching for artificial intelligence (AI) related data (or other high priority data). A host system may transmit a prefetch command for the AI-related data before performing one or more AI operations that use the data. In response, the memory system may prefetch the AI-related data and store the data in a cache. The storage of the AI-related data in the cache may be prioritized over other data types associated with lower priorities. The host system may transmit an FBO command to defragment the AI-related data in the memory system prior to the prefetch, which may reduce latency of the prefetch.

The present application relates to the field of cloud computing container platform, and discloses a container cluster resource fragmentation arrangement method, device, equipment and storage medium, the method comprises collecting node multidimensional resource utilization index data, and constructing a dynamic coloring model;The multidimensional resource utilization index data is input into the dynamic coloring model to obtain the node fragmentation index calculated by the model;Get the node fragmentation index threshold, compare the node fragmentation index with the node fragmentation index threshold, screen out the initial node to be arranged, calculate the arrangement investment return rate of each initial node to be arranged, select the node with a return rate greater than 1 as the target node to be arranged;Determine the arrangement window period;In the arrangement window period, the Pod migration problem of the target node to be arranged is modeled as a multidimensional packing problem, and the to-be-migrated Pod is obtained through an optimization algorithm, and the target node to be arranged is arranged based on the minimum migration cost principle.The present application solves the problem of resource fragmentation recovery lag and low cluster utilization in the prior art.

This application discloses a defragmentation method, device, and medium for GoldenDB databases, relating to the field of database data processing technology. The method includes: obtaining metadata information of all tables within the cluster from the management node of the GoldenDB distributed database; filtering tables that meet the defragmentation conditions from all tables based on preset dual thresholds to accurately determine the defragmentation targets; intelligently grouping the tables to be defragmented according to a preset defragmentation strategy and generating corresponding defragmentation tasks based on the grouping results; and distributing the defragmentation tasks to the computing nodes of the GoldenDB database for execution to complete the data defragmentation operation within the cluster. Through parallel execution by grouping, the total defragmentation time is reduced compared to existing random parallel schemes. The dual threshold filtering based on table size and fragmentation rate ensures that only tables requiring defragmentation are processed, saving system resources.

Techniques are provided for implementing a defragmentation process during a merge operation performed by a re-compaction process upon a log structured merge tree. The log structured merge tree is used to store keys of key-value pairs within a key-value store. As the log structured merge tree fills with keys over time, the re-compaction process is performed to merge keys down to lower levels of the log structured merge tree to re-compact the keys. Re-compaction can result in fragmentation because there is a lack of spatial locality of where the re-compaction operations re-writes the keys within storage. Fragmentation increases read and write amplification when accessing the keys stored in different locations within the storage. Accordingly, the defragmentation process is performed during a last merge operation of the re-compaction process in order to store keys together within the storage, thus reducing read and write amplification when accessing the keys.

This document describes methods, systems, and devices for handling increased file loads using file-based optimization (FBO) and prefetching within a storage system. A system can support the use of a prefetch command to prioritize the prefetching of artificial intelligence (AI)-related data (or other high-priority data). A host system can transmit a prefetch command for the AI-related data before performing one or more AI operations that utilize the data. In response, the storage system can prefetch the AI-related data and store it in a cache. Storing the AI-related data in the cache can be prioritized over other, lower-priority data types. The host system can also transmit an FBO command to defragment the AI-related data in the storage system prior to prefetching, which can reduce prefetching latency.

The full-disk defragmentation triggering method and related device provided by the embodiments of the present application relate to the technical field of terminals. The method comprises the following steps: determining whether to trigger the execution of FDD according to the IO state of the UFS device, the remaining life of the UFS device, and the time interval from the last execution of FDD, and dynamically calculating the interval time required for the next execution of FDD. In this way, the execution of FDD can be triggered on demand according to the state of the UFS device, such as the IO state and the remaining life, so that the freezing of the electronic device can be reduced, and the wear and tear on the UFS device can be reduced.

The application discloses a storage fragment processing method and device based on a cloud storage system, equipment and a medium, relates to the technical field of cloud computing and data processing, and can be applied to the fields of finance and medicine and comprises the following steps: acquiring multidimensional feature data of each storage fragment in a target storage space; determining the fragment types to which each storage fragment belongs based on the multidimensional feature data of each storage fragment; generating a target arrangement strategy for each fragment type according to the fragment types to which each storage fragment belongs; and performing an arrangement operation on each storage fragment in the target storage space according to the generated target arrangement strategy. The scheme effectively solves the problem that the arrangement effect is poor due to the use of a single threshold arrangement strategy in the prior art, and improves the arrangement effect of storage fragments.

The present disclosure provides a memory management method, device, electronic equipment, storage medium and program product, the method comprising: based on the task criticality level model, dividing the physical memory into several physically isolated memory regions, and setting differentiated access permissions for different memory regions; based on the task criticality level model, pre-allocating memory for tasks in the memory region corresponding to the corresponding task criticality level; monitoring the memory fragmentation distribution state of the task running to obtain a fragmentation entropy value, and in response to the fragmentation entropy value being greater than a fragmentation entropy value threshold, performing differentiated fragmentation consolidation strategies on different memory regions; predicting the memory pressure based on a load prediction model, and in response to the predicted memory pressure being greater than a memory pressure threshold, performing data migration and physical memory expansion; in response to the completion of task execution, recycling the memory resources occupied by the task, and based on the fragmentation entropy value, optimizing the fragmentation entropy value threshold and the memory pressure threshold. The strict real-time performance of high-priority real-time tasks is ensured.